ClinicBot: AI Chatbot for Guideline-Grounded Clinical Diagnosis
A team of researchers has created ClinicBot, an artificial intelligence tool aimed at delivering precise and verifiable clinical assistance by anchoring its responses in established medical protocols. This system tackles the hallucination issues prevalent in large language models, particularly in critical medical scenarios. ClinicBot features three major innovations: the structured extraction of clinical guidelines into semantic components (such as recommendations, tables, definitions, and narratives) with clear provenance, evidence prioritization that organizes information based on clinical relevance and guideline structure instead of mere textual similarity, and a web-based interface that offers succinct, actionable responses complete with verifiable citations. Detailed information about the system can be found in a paper available on arXiv (2605.00846).
Key facts
- ClinicBot is an AI system for clinical diagnosis support.
- It uses retrieval-augmented generation (RAG) with prioritized evidence.
- It extracts guidelines into semantic units: recommendations, tables, definitions, narrative.
- Evidence prioritization ranks by clinical significance, not textual similarity.
- The system provides verifiable citations for its answers.
- It aims to reduce hallucination in large language models for medical use.
- The paper is available on arXiv with ID 2605.00846.
- The system includes a web-based interface for concise, actionable answers.
Entities
—